BlueBench - RAG General: leaderboard
Metric: Score (%). Source: huggingface.co. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 Mini | 54.58 |
| 2 | Mistral Medium 3 | 52.57 |
| 3 | Granite 3.3 8B Instruct | 51.92 |
| 4 | Mistral Large | 51.08 |
| 5 | GPT-4.1 Nano | 50.88 |
| 6 | Llama 3.3 70B Instruct | 49.02 |
| 7 | GPT-4o | 48.9 |
| 8 | GPT-4.1 | 48.46 |
| 9 | Pixtral-12B | 47.85 |
| 10 | Llama 3.2 3B Instruct | 47.81 |
| 11 | Llama 3.2 1B Instruct | 44.7 |
| 12 | O3 Mini | 44.06 |
| 13 | O1 | 41.24 |
| 14 | O4 Mini | 40.32 |
Interactive version: theaggregate.ai/benchmark?slug=bluebench-rag-general · How It Works · Data refreshed daily, snapshot 2026-09-05.